Andrew Suffield dijo [Wed, Oct 22, 2003 at 04:53:30AM +0100]: > > I completely agree with you. A natural environment has a much > > larger probability to introduce mistakes than an artificial one - if a > > mistake appears when building in a overly limited artificial > > environment, we can quite confidently conclude that the packager > > omitted something. Most (trivial) FTBFS bugs I have inspected arise > > from an omitted build-dependency. Many, as Sven Luther points out, are > > Be very wary of listening to Sven Luther. His comments are frequently > disconnected from reality. ...I have quite a good impression from him - and even more on this particular thread: He is quite an expert regarding working on/for multiple architectures. As for me, OTOH, I am just a Perl programmer, most of my packages are arch-independent. And they are not autobuilt at all - I would love them to be autobuilt just to be sure my job did not get dirty from using other Perl modules. > I can't see how you can say this while "agreeing" with me. If we only > put artificially built packages in testing, then we are *not* testing > packages built in a real-world environment, so we have no real idea > how well they work. Maybe because my English is not native and I fail to understand some convoluted constructs? :) We *ARE* testing everything. If a package cannot be correctly built, in the rare case where this is not detected at build time, will fail miserably at its users - in Unstable or in Testing. This will help us ensure the test cycle is performed on the package quality itself. I fear that we have some (mainly architecture:all) packages in our stable release that FTBFS - which went undetected because no restricted, artificial builds were ever performed on it. Greetings, -- Gunnar Wolf - gwolf@gwolf.cx - (+52-55)5630-9700 ext. 1366 PGP key 1024D/8BB527AF 2001-10-23 Fingerprint: 0C79 D2D1 2C4E 9CE4 5973 F800 D80E F35A 8BB5 27AF
Attachment:
signature.asc
Description: Digital signature